The plan/activity/intention/goal recognition problem is the problem of identifying an agent’s intent by observing its behaviour. Traditionally, the problem has involved matching a sequence of observations to a plan in a pre-defined plan library, the winning plan being the one that "best" matches the observations. Recent developments, however, dispense with the overhead of a plan library and instead---based on the assumption that the observed agent is behaving rationally---take a cost-based approach using classical planning technology to generate cadidate plans as-needed over a model of the domain. In this talk, we will review this cost-based approach to goal recognition and some recent results both for general task planning and path planning. We will also pesent a preliminary framework of deception and techniques to handle irrational and deceptive agents when performing goal recognition.
This is joint work with Dr. Peta Masters and has appeared at AAMAS'17, IJCAI'17, JAIR'18 and AAMAS'19.